formatting NaiveDateTime as RFC3339 gives a bad-explaining error

[hasezoey]

Currently when trying to format a NaiveDateTime with RFC3339 (or to be more specific .format("%+")), there a very generic error:

thread 'main' panicked at 'a Display implementation returned an error unexpectedly: Error', /rustc/84c898d65adf2f39a5a98507f1fe0ce10a2b8dbc/library/alloc/src/string.rs:2533:14
stack backtrace:
   0: rust_begin_unwind
             at /rustc/84c898d65adf2f39a5a98507f1fe0ce10a2b8dbc/library/std/src/panicking.rs:579:5
   1: core::panicking::panic_fmt
             at /rustc/84c898d65adf2f39a5a98507f1fe0ce10a2b8dbc/library/core/src/panicking.rs:64:14
   2: core::result::unwrap_failed
             at /rustc/84c898d65adf2f39a5a98507f1fe0ce10a2b8dbc/library/core/src/result.rs:1750:5
   3: core::result::Result<T,E>::expect
             at /rustc/84c898d65adf2f39a5a98507f1fe0ce10a2b8dbc/library/core/src/result.rs:1047:23
   4: <T as alloc::string::ToString>::to_string
             at /rustc/84c898d65adf2f39a5a98507f1fe0ce10a2b8dbc/library/alloc/src/string.rs:2532:9
   5: bin::main
             at ./crates/bin/src/main.rs:10:10

no mention of that it cannot format because no timezone is provided

from my testing the error comes from:

chrono/src/format/mod.rs

Line 687 in 083d3bf

if let (Some(d), Some(t), Some(&(_, off))) = (date, time, off) {

because off is None and results in the following error:

chrono/src/format/mod.rs

Line 697 in 083d3bf

None => return Err(fmt::Error), // insufficient arguments for given format

versions:
rustc rustc 1.69.0 (84c898d65 2023-04-16)
chrono 0.4.24

reproduction code:

let test = NaiveDateTime::from_timestamp_millis(0).unwrap();
println!("test {}", test.format("%+"));

possible fixes:

• add a implicit zero timezone
• change the current error into something that has a message
• return a error before it even gets to this point explaining that RFC3339 is not supported for NaiveDateTime

[jtmoon79]

I was able to reproduce this issue on branch 0.4.x and branch main.

I added this user-provided regression test in PR #1097

[pitdicker]

#902 Should give a better error message here. It will have to wait until 0.5 though, as that is a breaking change.